• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Denafrips ARES II USB R2R DAC Review

Joined
Feb 27, 2020
Messages
12
Likes
0
"You realise any R2R DAC performing at 'modern' D-S performance levels, reaches it only via lots and lots of FPGA-programmed feedback, to offset the resistor errors throughout the conversion?"

Where does a resistor ladder consign only of resistors and switches have signal processing or feedback? Please explain.

Oversampling (bad name as it is simply a digital low pass filter) is of course a form of signal processing but that's independent of the type of DAC chip. We are comparing DS with R2R these are converters. Oversampling is another matter.
 

BDWoody

Major Contributor
Forum Donor
Joined
Jan 9, 2019
Messages
2,457
Likes
4,745
Location
Mid-Atlantic, USA. (Maryland)
"They typically haven't had very good execution...leading to very poor measured results."

So a PCM1704 has poor measured results?!?
Not sure how you made that leap, from me saying they typically haven't had good measured results, to you pulling out the exclamation points...but a quick search on this site will show some that have been tested, and typically it's just been sad.

Rather than start from scratch, in an off topic thread...maybe this would give more background on the general thinking.

https://www.audiosciencereview.com/...sigma-vs-“multibit”-what’s-the-big-deal.5359/
 

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
5,551
Likes
10,057
Location
The Neverlands
So a PCM1704 has poor measured results?!?
PCM1704 is not a 'typical' R2R based chip (it is sign magnitude) and a 20 year old chip. While it has very good results measurement wise it isn't quite reaching the level of the better DS applications.

Of course the final DAC device results not only depend on the used DAC chip but also how it used and what components/layout is used around the DAC chips and what the DAC chip is told to 'do'.
 
Joined
Feb 27, 2020
Messages
12
Likes
0
"'modern' D-S performance levels"

which are what?

Lets say 8 bit delta sigma?

The resistor ladder in these has a native resolution of 256 possible values or steps. The rest is apparent resolution through the use of signal processing and feedback.

A 24 bit ladder has a native resolution of 24 bit (proving the resistors can be trimmed precisely enough). No feedback or signal processing is present.

What is so difficult to understand?
 
Joined
Feb 27, 2020
Messages
12
Likes
0
"Whatever performance level is desired, a sigma-delta DAC can achieve it cheaper than an R-2R design. Either type can be made terrible, excellent, or somewhere in between."

Exactly what I intended with my original post. DS is cheaper. Not "better" as was implicated throughout the thread.
I am unaware of pricing but apparently a top notch DS like ESS sabre is apparently not exactly cheap.
 

BDWoody

Major Contributor
Forum Donor
Joined
Jan 9, 2019
Messages
2,457
Likes
4,745
Location
Mid-Atlantic, USA. (Maryland)
What is so difficult to understand?
I'm having a hard time understanding your point...

What am I missing.
It's harder to get r2r right, and there's no audible advantage over a $9 apple dongle.
 

mansr

Major Contributor
Joined
Oct 5, 2018
Messages
1,732
Likes
3,603
Location
Hampshire
A 24 bit ladder has a native resolution of 24 bit (proving the resistors can be trimmed precisely enough). No feedback or signal processing is present.
That's where you're mistaken. Resistor trimming alone can only take you so far. You then have to account for the remaining mismatch, drift, temperature gradients, etc. A couple of designs have managed to achieve very low distortion figures through (presumably individual) calibration and digital compensation. This takes a huge amount of engineering effort, and the resulting product is very expensive to make. Why do things the hard way?

Exactly what I intended with my original post. DS is cheaper. Not "better" as was implicated throughout the thread.
I am unaware of pricing but apparently a top notch DS like ESS sabre is apparently not exactly cheap.
Same performance cheaper is a very reasonable definition of better. The ES9038PRO costs $75 for samples which means it can probably be had for about $35 in 1k quantities. The Moolah-Moolah DAC has three ADI Sharc DSPs that cost about $20 each plus a bunch of other expensive parts.
 
Joined
Feb 27, 2020
Messages
12
Likes
0
"A couple of designs have managed to achieve very low distortion figures through (presumably individual) calibration and digital compensation."

Interesting. Which designs are that?

I understand the pcm1704 was the last R2R chip manufactured. There is no mention of digital signal processing in the datasheet.
 

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
5,551
Likes
10,057
Location
The Neverlands
The resistor ladder in these has a native resolution of 256 possible values or steps. The rest is apparent resolution through the use of signal processing and feedback.
You (conveniently) left out the important parts of the DS design.
The high frequency it operates on as opposed to the lower frequency of the R2R and the low pass filter.
 
Last edited:

Veri

Major Contributor
Joined
Feb 6, 2018
Messages
4,935
Likes
5,502
"A couple of designs have managed to achieve very low distortion figures through (presumably individual) calibration and digital compensation."
Interesting. Which designs are that?
Holo spring comes to mind. With the pinnacle being the Holo May. The thread topic DAC, the Denafrips probably uses a similar technique (although obviously less optimised). As do the Soekris boards via sign maltitude (digital) compensation.
 

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
5,551
Likes
10,057
Location
The Neverlands
There is no mention of digital signal processing in the datasheet.
Digital processing (over sampling for instance) is usually done before the DAC chip when using R2R based chips.
Using an R2R ladder NOS (which can be emulated using DS as well) is a very stupid thing to do so there goes the advantage of 'steps'.
In the end the analog output voltage is the goal.
Whether you create those DSD style, DS or other pulse bases or with analog sample-hold really does not matter when it comes to resolution.
It is a fallacy to think the generated voltages by a DAC device are more 'accurate' than those created by other techniques.
Yes, the out of band noise is smaller in R2R so it looks cleaner on a scope.
We listen to the signal not view it through a scope.
 
Joined
Feb 27, 2020
Messages
12
Likes
0
"Digital processing (over sampling for instance) is usually done before the DAC chip when using R2R based chips.
Using an R2R ladder NOS (which can be emulated using DS as well) is a very stupid thing to do so there goes the advantage of 'steps'.
In the end the analog output voltage is the goal.
Whether you create those DSD style, DS or other pulse bases or with analog sample-hold really does not matter when it comes to resolution.
It is a fallacy to think the generated voltages by a DAC device are more 'accurate' than those created by other techniques.
Yes, the out of band noise is smaller in R2R so it looks cleaner on a scope.
We listen to the signal not view it through a scope."

Agree that NOS does not really make sense although to me a good analog low pass filter sounds the same as a digital one albeit at higher cost.

True it is the output that counts.

Again: Throughout the thread it was claimed DS is better than R2R. A paper by Burr Brown claims otherwise.
The paper is old and with the newer DS like ESS sabre apparently the gap is closed. I never claimed one to be better. They are different in that DS is low bit, high speed, needing signal processing and feedback to obtain the required bit depth and the other not.
oversampling happens before the DAC and does not increase the bit depth of the DAC chip. If the chip is 24 bit than that is all it can do.
I still fail to see how starting with a lower bit DAC and using feedback and processing is magically better than than a native 24 bit DAC as claimed throughout the thread.
 
Joined
Feb 27, 2020
Messages
12
Likes
0
As an analogy:

If an amplifier with 0.01% THD can be built with transistors that are linear enough not require feedback i would prefer that over less linear transistors and lots of feedback.

I fail to see how using less linear transistors and a lot of feedback is better.

From a technical point of view I would prefer the first option as being more elegant and less complex.

Whether they sound the same I dunno but I'd sleep better:)
 

mansr

Major Contributor
Joined
Oct 5, 2018
Messages
1,732
Likes
3,603
Location
Hampshire
I sure hope you didn't mean "better sounding" ..? When starting this discussion. Or are you simply comparing the techniques.
Of course I didn't. If whatever requirement you have is met by several options, the cheaper alternative is generally considered better.
 

mansr

Major Contributor
Joined
Oct 5, 2018
Messages
1,732
Likes
3,603
Location
Hampshire
As an analogy:

If an amplifier with 0.01% THD can be built with transistors that are linear enough not require feedback i would prefer that over less linear transistors and lots of feedback.

I fail to see how using less linear transistors and a lot of feedback is better.

From a technical point of view I would prefer the first option as being more elegant and less complex.

Whether they sound the same I dunno but I'd sleep better:)
I guess you don't understand how amplifiers work either.
 
Top Bottom